修改列对象属性后Dataframe_Builder_Update未自动更新DataFrame问题

将由DataFrame列对象组成的数组传入Dataframe_Builder_Update类后，执行[setattr(obj, 'date', '12/29/2019') for obj in dataframe_builder_obj.column_builders]可成功更新列对象的日期及相关计算，但类的result_df不会自动同步更新。只有显式调用build_dataframe()方法才能得到正确结果，希望实现无需手动调用的自动更新。

原代码如下：

import pandas as pd

class Dataframe_Builder_Update():

    def __init__(self, column_builders):
        self._column_builders = column_builders
        self.build_dataframe()
    
    def build_dataframe(self):
        self.result_df = pd.DataFrame()

        for column_builder in self._column_builders:
            if not column_builder.group:
                self.result_df = pd.concat([self.result_df, column_builder.calculated_output], axis=0)
            elif column_builder.group:
                self.result_df = pd.concat([self.result_df, column_builder.calculated_output], axis=1)

    @property
    def column_builders(self):
        return self._column_builders

    @column_builders.setter
    def column_builders(self, new_column_builders):
        self._column_builders = new_column_builders
        self.build_dataframe() 


# 使用示例
# dataframe_builder_obj = Dataframe_Builder_Update(my_arr)
# dataframe_builder_obj.result_df

# 修改属性后result_df未更新
# [setattr(obj, 'date', '12/29/2019') for obj in dataframe_builder_obj.column_builders]
# dataframe_builder_obj.result_df

# 显式调用后正常
# [setattr(obj, 'date', '12/29/2019') for obj in dataframe_builder_obj.column_builders]
# dataframe_builder_obj.build_dataframe()

当前column_builders的setter仅在替换整个列对象数组时才会触发build_dataframe()更新result_df。但直接修改数组内单个列对象的属性时，主类无法感知到这些内部变化，因此不会自动重建DataFrame。

方案1：将result_df改为动态计算的属性 #

把result_df从实例变量改成@property，每次访问时自动重新构建。这种方式实现最简单，但如果build_dataframe()计算成本高，频繁访问会影响性能。

修改后的代码：

import pandas as pd

class Dataframe_Builder_Update():

    def __init__(self, column_builders):
        self._column_builders = column_builders
    
    def build_dataframe(self):
        result_df = pd.DataFrame()
        for column_builder in self._column_builders:
            if not column_builder.group:
                result_df = pd.concat([result_df, column_builder.calculated_output], axis=0)
            else:
                result_df = pd.concat([result_df, column_builder.calculated_output], axis=1)
        return result_df

    @property
    def column_builders(self):
        return self._column_builders

    @column_builders.setter
    def column_builders(self, new_column_builders):
        self._column_builders = new_column_builders

    @property
    def result_df(self):
        # 每次访问时自动构建最新的DataFrame
        return self.build_dataframe()

使用时，直接访问dataframe_builder_obj.result_df就能得到基于最新列对象的结果，无需手动调用build_dataframe()。

方案2：给列对象添加观察者机制 #

让列对象在自身属性变化时通知Dataframe_Builder_Update实例，触发更新。这种方式更高效，只有当列对象确实变化时才会重建DataFrame。

首先定义列对象的基类，添加属性变更通知：

import pandas as pd

class ColumnBuilder:
    def __init__(self, date, group=False):
        self._date = date
        self.group = group
        self._observers = []

    def add_observer(self, observer):
        self._observers.append(observer)

    def _notify_observers(self):
        for observer in self._observers:
            observer.update()

    @property
    def date(self):
        return self._date

    @date.setter
    def date(self, new_date):
        self._date = new_date
        self._notify_observers()  # 属性变更时通知观察者

    @property
    def calculated_output(self):
        # 示例计算逻辑，根据date生成DataFrame
        return pd.DataFrame({f'col_{self._date}': [1,2,3]})

然后修改Dataframe_Builder_Update类，实现观察者接口：

class Dataframe_Builder_Update():

    def __init__(self, column_builders):
        self._column_builders = column_builders
        # 给每个列对象添加当前实例作为观察者
        for builder in self._column_builders:
            builder.add_observer(self)
        self.build_dataframe()
    
    def update(self):
        # 收到列对象变更通知时，自动重建DataFrame
        self.build_dataframe()
    
    def build_dataframe(self):
        self.result_df = pd.DataFrame()
        for column_builder in self._column_builders:
            if not column_builder.group:
                self.result_df = pd.concat([self.result_df, column_builder.calculated_output], axis=0)
            else:
                self.result_df = pd.concat([self.result_df, column_builder.calculated_output], axis=1)

    @property
    def column_builders(self):
        return self._column_builders

    @column_builders.setter
    def column_builders(self, new_column_builders):
        # 移除旧列对象的观察者
        for builder in self._column_builders:
            if self in builder._observers:
                builder._observers.remove(self)
        self._column_builders = new_column_builders
        # 添加新列对象的观察者
        for builder in self._column_builders:
            builder.add_observer(self)
        self.build_dataframe()

使用时，修改列对象的date属性后，result_df会自动更新：

# 创建列对象
col1 = ColumnBuilder('12/28/2019')
col2 = ColumnBuilder('12/28/2019', group=True)
my_arr = [col1, col2]

# 初始化构建器
dataframe_builder_obj = Dataframe_Builder_Update(my_arr)
print(dataframe_builder_obj.result_df)

# 修改列对象日期，result_df自动更新
col1.date = '12/29/2019'
print(dataframe_builder_obj.result_df)

方案3：在主类中封装修改列属性的方法 #

不在外部直接修改列对象的属性，而是通过主类提供的接口来操作，这样主类可以在修改后立即触发更新。

修改后的Dataframe_Builder_Update类：

import pandas as pd

class Dataframe_Builder_Update():

    def __init__(self, column_builders):
        self._column_builders = column_builders
        self.build_dataframe()
    
    def build_dataframe(self):
        self.result_df = pd.DataFrame()
        for column_builder in self._column_builders:
            if not column_builder.group:
                self.result_df = pd.concat([self.result_df, column_builder.calculated_output], axis=0)
            else:
                self.result_df = pd.concat([self.result_df, column_builder.calculated_output], axis=1)

    @property
    def column_builders(self):
        return self._column_builders

    @column_builders.setter
    def column_builders(self, new_column_builders):
        self._column_builders = new_column_builders
        self.build_dataframe()

    def update_column_date(self, index, new_date):
        # 根据索引修改指定列对象的日期，并触发更新
        if 0 <= index < len(self._column_builders):
            setattr(self._column_builders[index], 'date', new_date)
            self.build_dataframe()

    def update_all_columns_date(self, new_date):
        # 修改所有列对象的日期，并触发更新
        for obj in self._column_builders:
            setattr(obj, 'date', new_date)
        self.build_dataframe()

使用时，通过主类的方法来修改日期：

# 初始化后，修改所有列的日期
dataframe_builder_obj.update_all_columns_date('12/29/2019')
# 此时result_df已经自动更新
print(dataframe_builder_obj.result_df)

内容的提问来源于stack exchange，提问作者steezebutter